A parametric simulator-generated event stream for 60 simulated restaurant kitchens across 7 days. Designed for benchmarking demand forecasting, operational-state inference, and self-supervised representation learning on time-series event data. This is the publish-ready version. A hidden ground-truth state column (hidden_true_state) was used during the original simulator's classifier and is stripped here — only what a… See the full description on the dataset page: https://huggingface.co/datasets/Vita-Mojo/kitchensim-v1.